Development of a Hi-Speed Near Real-Time 720i Image Processing Application for Flight Test

Transcription

1

2 Development of a Hi-Speed Near Real-Time 720i Image Processing Application for Flight Test Luiz Eduardo Guarino de Vasconcelos Nelson Paiva Oliveira Leite, PhD Carlos Alberto Murari Pinheiro, PhD Otávio Augusto Salgado Carpinteiro, PhD

3 Test Point Goal: Altitude and Airspeed Reference Points

4 Introduction Images frames are used as information source to clearly pinpoint the aircraft behaviour at the FTC. Air Data System (ADS) Calibration FTC using the tower-fly-by method requires the knowledge of the exact aircraft reference altitude. The IPEV uses an off-line video processing application that computes the aircraft altitude from a snap-shot picture. Main disadvantage: measurement accuracy is jeopardized. Solution: application to process 720i video frames at up to 400 fps to be used for ADS calibration FTC

5 Imaging Processing Brightness +143% and Contrast + 79%

6 Imaging Processing Many applications areas Fingerprint recognition Digital mammography Tempel-1 Comet Electronic Circuit

7 Essential to Flight Safety ADS FTC Overview Derived from Dynamic and Static Pressures Computation requires Calibration in Laboratory and FTC

8 ADS Calibration Consists n TP: 1.2 Stall Speed to VH Requirements for Valid Test Point: Z pb and V b should be stabilized Where: maximum deviation of basic speed at the i th (kts); maximum deviation of the aircraft altitude at the i th (ft); mean basic speed at the i th test point (kts) scheduled basic speed for the i th test point (kts);

9 ADS Calibration Valid Test Point with H-55 Helicopter Requirements are considered within the valid area (ARV) Reference Points in the lower corners

10 ADS Calibration Test Site ADS Mantain Trajectory Camera and RP are static and known position

11 Challenges Automatic detection 6 x 6 pixels

12 Challenges Weather Conditions

13 Tool Development User makes application setup Video frames transmitted Extraction of target coordinates Correction to minimize errors lens distortion Computes TSPI Results

14 Algorithms: Tool Development Reference point detection; Tarmac centreline detection; Aircraft detection; TSPI Computing; and Test Point Validation

15 Aircraft Detection Segmentation First Frame TP is Reference Background Image, must have aircraft and pepper noises

16 Aircraft Detection Next steps: Detect edges Perform CCL Remove pepper noises Sort ascending size order (CCL) Biggest CCL contains aircraft Perimeter Pixels pepper noises

17 TSPI Computing Now, requirement is find a fixed reference point on the aircraft (RPA) for measure altitude and airspeed Centroid, Front, Rear and Bottom Edge detection After several tests, the Rear is the better Found PRA, computes altitude and airspeed PRA PR #1 Tarmac Centreline PR #2

18 Tool Evaluation HELIBRAS Esquilo H55 MacroVis SpeedCam MatLab environment Intel Pentium IV Core 2 Duo CPU T GHz notebook, 4 Gb RAM and Microsoft Windows 7 Professional. Camera: best configuration was to generate images in grayscale, 400 fps and 720i resolution.

19 Tool Evaluation Runs at 48 fps ± 2 Single frame of a valid TP

20 Tool Evaluation True Reference and Computed Altitude

21 Tool Evaluation Altitude Uncertainty

22 Summary The development and evaluation of a Hi-Speed near Real-Time 720i Image Processing Application for Flight Test was successfully executed. This application integrates several simple yet efficient vision tools, which are easy to implement. The system can be customized for several aircrafts. As a result the system is very flexible and reliable and it can be used in wide range applications.

23 Summary The next steps are: Evaluate the tool with other aircrafts; Improve system performance using: Parallel processing techniques; and Graphics Processor Unit (GPU) cards; Retrieve images directly from the buffer of high-speed camera; and Develop a tool to integrate this application with GPS and GTS.

24 Acknowledgement We wish to thank the partial support given by the Flight Test Research Institute, specially the Flight Test Course Students, for supporting the measurement and the ADS calibration flight tests campaigns. Also we like to thank FINEP under agreement that funded the development of this tool and the presentation trip.